[llvm] AMDGPU/GlobalISel: Fix inst-selection of ballot (PR #109986)

[Matt Arsenault via llvm-commits]

================
@@ -89,7 +90,8 @@ define amdgpu_cs i32 @branch_divergent_ballot_ne_zero_non_compare(i32 %v) {
 ; CHECK:       ; %bb.0:
 ; CHECK-NEXT:    v_and_b32_e32 v0, 1, v0
 ; CHECK-NEXT:    v_cmp_ne_u32_e32 vcc_lo, 0, v0
-; CHECK-NEXT:    s_cmp_eq_u32 vcc_lo, 0
+; CHECK-NEXT:    s_and_b32 s0, vcc_lo, exec_lo
----------------
arsenm wrote:

The output was a real compare, so this shouldn't be needed. The DAG doesn't emit this and 

https://github.com/llvm/llvm-project/pull/109986